Study design
The study is a prospective, double-blinded randomized placebo-controlled pilot feasibility trial in a pediatric cardiac ICU at Boston Children’s Hospital, a quaternary free-standing children’s hospital. Enrollment began in February 2019 with a planned enrollment of 100 patients over a 2-year period. The local Institutional Review Board (IRB) at Boston Children’s Hospital approved this study. All study protocol amendments, deviations or adverse events will be immediately reported to IRB. All research team members, clinicians, data analysts, and trial participants are blinded to study assignments. Unblinding will only occur in the event a participant has a serious adverse event such a clinically significant gastrointestinal bleed. The study was registered at ClinicalTrials.gov (NCT03667703) and funded by The Gerber Foundation’s National Research Grant (#5781). The funding agency will have no access to or involvement in the data analysis or writing of the manuscripts. The research integrity, data quality, and adverse event assessment will be regularly reviewed by a pre-appointed, independent Data Safety Monitoring Board (DSMB). The DSMB has an appointed chair and five other members. Further details regarding the charter can be available by contacting the corresponding author who is a co-principal investigator. Table 1 shows the Standard Protocol Items: Recommendation for Interventional Trials (SPIRIT) schedule for the enrollment, intervention and assessment periods. The SPIRIT checklist is in Additional file 1.
Table 1
Standard Protocol Items: Recommendation for Interventional Trials (SPIRIT) schedule of enrollment, interventions and assessments.
| | Study Period |
Enrollment | Allocation | Post-allocation | Close-out |
Timepoint | Pre-study | Study day #1 | Study day #2 through up to 14 | 24 hours after respiratory support discontinued | Hospital discharge |
Enrollment | Eligibility screen | υ | υ | | | |
Informed consent | υ | υ | | | |
Allocation | | υ | | | |
Intervention | H2 blocker or placebo | | υ | υ | | |
Assessments | Demographics | | υ | υ | υ | |
| Laboratory data | | υ | υ | υ | |
| Antibiotic exposure | | υ | υ | υ | |
| Inotrope and vasoactive use | | υ | υ | υ | |
| Nutritional support | | υ | υ | υ | |
| Gastrointestinal bleeding | | υ | υ | υ | υ |
| Necrotizing enterocolitis | | υ | υ | υ | υ |
| Infectious complications | | υ | υ | υ | υ |
| Adverse event | | υ | υ | υ | υ |
| Oral, gastric, blood and urine samples | | υ | | υ | υ |
| Stool samples | | υ | υ | υ | υ |
| Length of stay | | | | | υ |
| Mortality | | | | | υ |
Study population
The inclusion and exclusion criteria are described in Table 2. Infants
Table 2
Inclusion criteria | (1) < 12 months of age (including premature newborns), (2) diagnosed with congenital heart disease*, (3) admitted to the CICU, (4) anticipated to require respiratory support¶ for > 24 hours during their CICU stay, and (5) have received ≤ 1 dose of stress ulcer prophylaxisβ during their current admission. |
Exclusion criteria | (1) prior use of antacidsβ in the past month for > 7 days, (2) active gastrointestinal bleeding, (3) active Helicobacter pylori infection, (4) anticipated exposure to certain pharmaceuticals: (i) high-dose steroids (equivalent to 4 mg/kg/day of methylprednisolone), (ii) intravenous non-steroidal anti-inflammatory drugs (ie. ketorolac), (iii) certain anticoagulants including high-dose aspirin, direct thrombin inhibitors and GPIIbIIIa inhibitors, (5) planned to undergo or recently has undergone gastrointestinal surgery within the last 4 weeks, (6) supported by ECMO or VAD, (7) currently enrolled in another conflicting interventional trial, (8) known to be allergic to H2RAs, (9) admitted for palliative care, (10) prior enrollment in the study, OR (11) primary provider declines enrollment. |
*CHD includes anatomic, myopathic and arrhythmic conditions |
¶Respiratory support includes mechanical ventilation, non-invasive positive pressure ventilation, and high-flow oxygen therapy |
βStress ulcer prophylaxis or antacids includes H2RAs, PPIs, and sucralfate. |
CICU, cardiac intensive care unit; H2RA, histamine-2 receptor antagonist; PPI, proton pump inhibitor; GIIbIIIa, glycoprotein IIb/IIIa; ECMO, extracorporeal membrane oxygenator; VAD, ventricular assist device |
diagnosed with CHD admitted to the cardiac ICU and anticipated to require respiratory support (defined below) for greater than 24 hours will be eligible for the study. Congenital heart disease includes anatomic, myopathic, and arrhythmic conditions. Respiratory support is defined as mechanical ventilation, including conventional, high frequency oscillatory, or jet ventilation, as well as non-invasive positive pressure ventilation, such as continuous (CPAP) and biphasic (BIPAP) positive airway pressure, and high-flow nasal cannula. Respiratory support for greater than 24 hours was chosen as a surrogate for severity of illness. Patients will be excluded if they receive any form of antacid for > 7 days during the past month as this could potentially alter their baseline gut microbiome. Patients will also be excluded if they are anticipated to receive high-dose steroids, intravenous non-steroidal anti-inflammatory agents, or high-dose aspirin during their hospitalization, as these medications may potentially cause gastritis and increase the risk for UGI bleeding. Finally, infants on certain anticoagulants – direct thrombin inhibitors and GPIIbIIIa inhibitors – will be excluded since these medications do not have an available reversal agent in the event of an UGI bleed.
Recruitment and study flow
Eligible patients will be screened for eligibility and one of the principal investigators will approach the family of an eligible patient for written consent. If the legal guardian grants consent, the patient will be enrolled in the study and then randomized to one of two arms. Study procedures will be continued until the patient: (1) no longer requires respiratory support for greater than 24 hours, (2) transfers to the floor or is discharged from the cardiac ICU, (3) completes 14 days of study drug, or (4) at any time during the study the primary provider believes that open-label acid suppression is indicated (Fig. 1).
Randomization
Eligible patients will be randomly assigned by Boston Children’s Research Pharmacy to receive either a histamine-2 receptor antagonist (i.e. ranitidine or famotidine per institutional standard) or placebo. The randomization assignments are generated by Boston Children’s Hospital’s proprietary randomization software, SciRan®. Randomization will be performed within 2 strata defined by admission type (medical or surgical) and age (neonate, age < 30 days, or infant, 1 month to 1 year). Allocation will be using permuted blocks in a 1:1 ratio to ensure balanced allocations across the two treatment groups within each stratum.
Sample size
We powered the precision of our feasibility estimates for both screening and drug initiation – which we deemed as the two most important feasibility measures. There are approximately 600 patients under one year of age admitted to the cardiac ICU each year, and we anticipate that an estimated 200 patients will be eligible over the two-year recruitment period. The target to demonstrate feasibility is screening 80% of all patients. If n = 600, the lower limit of the 95% one-sided confidence interval will include 80% as long as the observed screening rate is at least 82.7%. There are 100 patients to be randomized in this pilot trial. The target to demonstrate feasibility with respect to drug initiation is to have 80% of randomized patients receive their first dose of study drug within 48 hours. With n = 100, the lower limit of the 95% one-sided confidence interval will include 80% as long as the observed drug initiation rate is at least 86.9%. That is, as long as the observed drug initiation rate is at least 86.9%, we can be 95% confident that the drug initiation rate, to be realized in a future trial, is at least 80%. The study is not powered to assess a statistical difference between the incidence of UGI bleeding and hospital-acquired infections, as the historical incidences are very low, 0.5% and 2%, respectively. These outcomes will be further assessed in a future larger, multi-center trial.
For the assessment of oral, gastric, and stool microbiota, a total of 600 samples (2 samples per site) will be obtained from the 100 patients. We estimate detectable effect sizes using these sample sizes based on the Human Microbiome Project (14). With adjustment for three covariates: one categorical binary (antibiotics), one continuous (age) and one categorical tertiary (hospital-acquired infections), we anticipate a power of 0.9 to detect a 1.36% level of rare taxon relative abundance and a 3.46% level of common taxa relative abundance, within a treatment arm between two time points, and similarly between two treatment arms at a given time point for either oral, gastric, or stool samples. All estimates incorporate stringent Bonferroni multiple-hypothesis correction to an adjusted p-value of 0.05. Approximating as above, the corresponding detection limit for 250 metagenomic pathways with power of 0.9 would be 3.87%.
Study intervention
Study participants will be randomly assigned to receive a histamine-2 receptor antagonist (i.e. ranitidine or famotidine) or placebo. As there are no pediatric-specific recommendations regarding type of SUP, we elected to conduct our study with a histamine-2 receptor antagonist since this is the current standard practice at our institution. In this pragmatic design, participants can receive study drug either intravenously or enterally, depending on the clinician’s preference. The dosing is based on age and route. For ranitidine, neonates (< 31 days old) will receive either 1 mg/kg intravenously/enterally every 12 hours, and infants (≥ 31 days old) will receive 1 mg/kg intravenously/enterally every 8 hours. For famotidine, if the participant is < 3 months they will receive 0.5 mg/kg/dose IV/PO daily and if > 3 months old they will receive 0.25 mg/kg/dose IV every 12 hours or 0.5 mg/kg/dose PO every 12 hours.
The placebo will be in an equivalent volume of 0.9% saline intravenously or a stevioside sugar-free syrup vehicle that resembles the color, tonicity, and texture of oral ranitidine or famotidine. Individualized unit-dose syringes will be provided to each study participant and blinded to the study team, bedside clinicians, parents, and outcome assessors. All interventions other than the study drug are left to the discretion of the treating clinicians. Once a participant completes the study, the clinicians can prescribe acid suppression per their usual practice. Adherence to intervention protocols is monitored daily by research staff by checking in with participant’s bedside nurse to ensure medication administration and timely sample collection. Research staff also discuss participant with attending physician daily to assess for any adverse events or study related issues. Patients with hemodynamically significant UGI bleeding will be withdrawn from the study, as they would likely require open-label acid suppression. There also exists study-halting criteria, which would stop the study until the DSMB reviewed and recommended continuation of the study. Final determination of trial termination will be made by principal investigators, if necessary. Auditing of trial conduct is done every six months by the DSMB, independent of the investigators and sponsor. The study-halting criteria include a total of 3 UGI bleeding events or an enrollment number of less than 20 patients per year. The principal investigators will have access to the final trial dataset.
Measurements and definitions
Data will be imported into a secure, password-protected, FDA-compliant database (InForm® Electronic Data Capture). Data collection will include demographic, procedural, laboratory, pharmaceutical, nutritional, ventilatory and outcome variables. Samples obtained will include oral swabs, gastric aspirates (via indwelling NG tube), discarded blood, and urine at the initiation and conclusion of the study. In addition, serial stool samples will be collected while on study.
Important definitions include the following:
(1) Clinically significant UGI bleed: new-onset bleeding from the UGI tract (i.e. hematemesis, bloody gastric aspirate, or hematochezia) AND associated with: (a) decrease in hemoglobin by 2 g/dL, OR (b) decline in mean arterial blood pressure by 10 mmHg or initiation/increase of inotrope/vasoactive medications, OR (c) increase in heart rate by 20 beats per minute in the absence of an arrhythmia or fever, OR (d) need for unanticipated blood transfusion, OR (e) unexpected endoscopic or operative procedure to achieve hemostasis. This definition has been used in adult randomized controlled trials with excellent inter-rater agreement (15, 16).
(2) Bloodstream infection (BSI): a laboratory-confirmed bloodstream infection with or without a central line in place.
(3) Ventilator associated event (VAE): a deterioration in respiratory status after a period of stability or improvement on the ventilator, evidence of infection or inflammation and laboratory evidence of a respiratory infection (CDC).
(4) Urinary tract infection: a laboratory-confirmed urinary tract infection with or without a urinary catheter in place.
(5) Clostridium difficile associated diarrhea: Diarrhea in the presence of a positive C. difficile test.
(6) Mediastinitis: a laboratory-confirmed organism from mediastinal tissue or fluid, or based on gross anatomic exam, or has hyper/hypothermia or apnea or bradycardia or sternal instability with at least one of the following: (a) purulent drainage from the mediastinal area or (b) mediastinal widening on imaging (CDC).
(7) Superficial wound infection: has two of the following symptoms: (a) erythema, (b) tenderness, (c) swelling AND a laboratory-confirmed organism is identified from the wound.
Sample analyses
For assessment of stool microbiota, sequence-based microbial community surveys of stool samples will be carried out by 16S rRNA gene-based sequencing in the Microbial Genomics and Transcriptomics Core at the Broad Institute. Their protocol targets the V4 window of the 16S rRNA gene and uses Illumina MiSeq system at the Broad Genomics Platform to produce on average 25,000 quality filtered, stitched paired-end reads per sample, representing the current state-of-the-art (17) In addition, a higher resolution community survey and complementary functional survey of a subset of the stool samples (20%) will be obtained using metagenomic shotgun DNA sequencing, to be performed at the Broad Institute in the Broad Technology Labs and Genomics Platform. Metagenomic libraries will be constructed using the Nextera XT DNA library preparation kit (Illumina) and sequencing will be performed on an Illumina HiSeq 2000 to generate a minimum of 2 Gb of 101 nt paired end reads.
For evaluation of oral and gastric microbiota, samples will be subjected to ribosomal DNA (rDNA) amplicon sequencing to characterize the composition of these microbiome communities. First, DNA will be extracted using a robust commercial extraction kit. Next, rDNA hypervariable regions of specific kingdoms will be amplified using universal primers. For bacteria, 16S V3V4 will be targeted using 5'-CCTACGGGNGGCWGCAG-3' and 5'-GGACTACNVGGGTWTCTAAT-3'; for fungi, ITS1 is targeted using 5'-CTYGGTCATTTAGAGGAAGTAA-3' and 5'-GCTGCGTTCTTCATCGATGC-3'. Specifically, Illumina adapter sequences and variable spacers are added 5’ upstream of these sequences to enable high-throughput multiplexed sequencing. The prepared amplicons will then be pooled and sequenced using Illumina Miseq 2 × 300 bp platform. Illumina raw reads will be de-multiplexed, quality trimmed, dereplicated and denoised, and finally mapped against established rDNA databases. The derived OTU table will document the relative abundance of each taxonomy within each sample.
Outcomes
The central objective for our pilot study is to investigate whether a clinical trial assessing the safety and efficacy of withholding SUP in infants with CHD in the cardiac ICU is feasible. The trial will be considered feasible if each of the following 4 a-priori variables are met: (1) > 80% of eligible patients are approached for consent (screening), (2) > 20% of eligible patients are randomized (enrollment and consent), (3) > 80% of consented patients received their first dose of study drug within 48 hours (allocation) and (4) > 80% protocol compliance achieved (protocol adherence). In addition to feasibility, we will assess safety by comparing the difference in the incidence of clinically significant UGI bleeding and hospital-acquired infections between participants receiving SUP versus placebo. Finally, we will explore the changes to the gut microbiota by comparing the absolute and serial differences in the abundance of bacteria and functional microbial profiles between those receiving SUP compared to placebo.
Statistics
The study will be double-blinded and we will utilize an intention-to-treat (primary analysis) and per protocol (secondary) analysis. Trial participants who do not complete the intervention will remain in the primary, intention-to-treat analysis of trial outcomes. Data will be reviewed every 6 months by the DSMB. Homogeneity of the two treatment arms will be assessed using a Fisher exact or Chi-square test for categorical variables and a Student t test (parametric) or Wilcoxon rank sum test (non-parametric) for continuous variables.
Feasibility analysis: For each feasibility outcome measure we will report the proportions of screened patients and/or participants meeting each criterion successfully and the associated one-sided 95% confidence interval.
Safety analysis: A two-sided 95% confidence interval will be constructed for treatment difference in the proportion of patients with UGI bleeding and hospital-acquired infections, as well as for each treatment-arm specific rate of UGI bleeding.
Microbiome analysis: Once the rRNA sequencing is completed, we will perform taxonomic profiling to identify distinct microbial lineages and then compare them to the published Greengenes, SILVA (for 16S) and UNITE (for ITS1) Reference Database (18–20). The primary characteristics to be assessed between treatment arms are: (a) within-sample and between-sample overall ecology of the microbial community, (b) absolute and relative abundance of microbial communities, and (c) pattern classification analysis to identify diversity (21). We will then perform per-feature multivariable association analyses that estimate which microbiome attributes differ by treatment arm as well as between time points, while accounting for covariates, to identify how microbes are affected by outcomes in the presence of certain covariates (22–24).